FAIR² Access Governance Documentation

Level 6 – Federated Compute-to-Data Access

This document provides the case scenario matrix for Level-6 access, together with metadata examples for each variant.
It follows the FAIR² interpretation:

Level 6 – Federated Compute-to-Data Access
Data cannot be downloaded.
Algorithms are sent to the data (“compute-to-data”).
Strict identity, authorization, environment, and export rules apply.
Only aggregated or model-derived outputs may leave the secure compute environment.
Suitable for highly sensitive, multi-institutional, cross-jurisdiction datasets.

1. Overview of Level 6 (Federated Compute-to-Data)

Level 6 introduces the strongest technical controls short of Level 5’s secure human environment.
The key characteristic is:

❗ No raw data leaves the sovereign hosting environment.

FAIR² models Level 6 using these constraint axes:

identity: ORCID / institutionalLogin / both
environment: computeToData (primary), hybrid if mixtures exist
communityGovernance: none / consultationRequired / communityApprovalRequired
aiTraining: allowed / restricted / prohibited
exportRestriction: derived-parameters-only / aggregated-only
distribution-level constraints: for multi-file governance

2. Level 6 Case Scenario Matrix

Variant	Description	identity	environment	governance	aiTraining	exportRestriction	distribution-level constraints
6.1	Pure federated compute-to-data	ORCID	computeToData	none	allowed	derived-parameters-only	none
6.2	Federated compute + restricted AI training	ORCID	computeToData	none	restrictedWithConditions	aggregated-only	none
6.3	Federated compute + AI training prohibited	ORCID	computeToData	none	prohibited	aggregated-only	none
6.4	Federated compute + community consultation	ORCID	computeToData	consultationRequired	restrictedWithConditions	aggregated-only	none
6.5	Federated compute + community approval required	ORCID	computeToData	communityApprovalRequired	prohibited	aggregated-only	none
6.6	Hybrid dataset (some files open, core data compute-only)	ORCID	hybrid	consultationRequired	allowed	derived-parameters-only	present

3. Metadata Examples for Each Level-6 Variant

All examples use the FAIR² JSON-LD namespace:

fair2: https://fair2.ai/ns/

Variant 6.1 — Pure Federated Compute-to-Data

{
  "@context": {
    "schema": "https://schema.org/",
    "dct": "http://purl.org/dc/terms/",
    "fair2": "https://fair2.ai/ns/"
  },
  "@type": "schema:Dataset",
  "schema:name": "Hospital Network Readmission Dataset (Federated Access)",
  "dct:license": "https://opendatacommons.org/licenses/by/1-0/",
  "dct:accessRights": "fair2:AgreementLevel6",
  "fair2:constraints": {
    "identity": "ORCID",
    "environment": "computeToData",
    "communityGovernance": "none",
    "aiTraining": "allowed",
    "exportRestriction": "derived-parameters-only"
  },
  "fair2:reuseLogging": true
}

Variant 6.2 — Federated Compute with Restricted AI Training

{
  "@context": {
    "schema": "https://schema.org/",
    "dct": "http://purl.org/dc/terms/",
    "fair2": "https://fair2.ai/ns/"
  },
  "@type": "schema:Dataset",
  "schema:name": "Cross-Institutional Laboratory Data (Federated, Restricted AI Use)",
  "dct:license": "https://opendatacommons.org/licenses/by/1-0/",
  "dct:accessRights": "fair2:AgreementLevel6",
  "fair2:constraints": {
    "identity": "ORCID",
    "environment": "computeToData",
    "communityGovernance": "none",
    "aiTraining": "restrictedWithConditions",
    "exportRestriction": "aggregated-only"
  },
  "fair2:reuseLogging": true
}

Variant 6.3 — Federated Compute, AI Training Prohibited

{
  "@context": {
    "schema": "https://schema.org/",
    "dct": "http://purl.org/dc/terms/",
    "fair2": "https://fair2.ai/ns/"
  },
  "@type": "schema:Dataset",
  "schema:name": "Sensitive Genomic Variation Set (Federated, No Training)",
  "dct:license": "https://opendatacommons.org/licenses/by/1-0/",
  "dct:accessRights": "fair2:AgreementLevel6",
  "fair2:constraints": {
    "identity": "ORCID",
    "environment": "computeToData",
    "communityGovernance": "none",
    "aiTraining": "prohibited",
    "exportRestriction": "aggregated-only"
  },
  "fair2:AIethicalUse": "Output limited to aggregated statistics; no AI training permitted.",
  "fair2:reuseLogging": true
}

Variant 6.4 — Federated Compute + Community Consultation Required

{
  "@context": {
    "schema": "https://schema.org/",
    "dct": "http://purl.org/dc/terms/",
    "fair2": "https://fair2.ai/ns/"
  },
  "@type": "schema:Dataset",
  "schema:name": "Traditional Knowledge Biodiversity Database (Federated Access)",
  "dct:license": "https://opendatacommons.org/licenses/by/1-0/",
  "dct:accessRights": "fair2:AgreementLevel6",
  "fair2:constraints": {
    "identity": "ORCID",
    "environment": "computeToData",
    "communityGovernance": "consultationRequired",
    "aiTraining": "restrictedWithConditions",
    "exportRestriction": "aggregated-only"
  },
  "fair2:governanceProtocol": "https://community.example.org/biodiversity-governance",
  "fair2:reuseLogging": true
}

Variant 6.5 — Federated Compute + Community Approval Required

{
  "@context": {
    "schema": "https://schema.org/",
    "dct": "http://purl.org/dc/terms/",
    "fair2": "https://fair2.ai/ns/"
  },
  "@type": "schema:Dataset",
  "schema:name": "Indigenous Genomics Panel (Federated + Approval)",
  "dct:license": "https://opendatacommons.org/licenses/by/1-0/",
  "dct:accessRights": "fair2:AgreementLevel6",
  "fair2:constraints": {
    "identity": "ORCID",
    "environment": "computeToData",
    "communityGovernance": "communityApprovalRequired",
    "aiTraining": "prohibited",
    "exportRestriction": "aggregated-only"
  },
  "fair2:governanceProtocol": "https://community.example.org/genomics-governance",
  "fair2:reuseLogging": true
}

Variant 6.6 — Hybrid Dataset: Public Metadata + Federated Raw Data

{
  "@context": {
    "schema": "https://schema.org/",
    "dct": "http://purl.org/dc/terms/",
    "fair2": "https://fair2.ai/ns/"
  },
  "@type": "schema:Dataset",
  "schema:name": "Urban Mobility + Environmental Sensors (Hybrid Federated Access)",

  "dct:license": "https://opendatacommons.org/licenses/by/1-0/",
  "dct:accessRights": "fair2:AgreementLevel6",

  "fair2:constraints": {
    "identity": "ORCID",
    "environment": "hybrid",
    "communityGovernance": "consultationRequired",
    "aiTraining": "allowed",
    "exportRestriction": "derived-parameters-only"
  },

  "distribution": [
    {
      "@type": "schema:DataDownload",
      "schema:contentUrl": "https://example.org/files/open_air_quality_summary.csv",
      "fair2:constraints": {
        "environment": "open"
      }
    },
    {
      "@type": "schema:DataDownload",
      "schema:contentUrl": "https://secure.example.org/federated/mobility_data",
      "fair2:constraints": {
        "identity": "ORCID",
        "environment": "computeToData",
        "exportRestriction": "derived-parameters-only"
      }
    }
  ],

  "fair2:governanceProtocol": "https://community.example.org/sensor-governance",
  "fair2:reuseLogging": true
}

4. Summary Table

Variant	Meaning	Key FAIR² Constraint
6.1	Pure compute-to-data	`environment = computeToData`
6.2	Federated + limited AI training	`aiTraining = restrictedWithConditions`
6.3	Federated + no AI training	`aiTraining = prohibited`
6.4	Federated + community consultation	`communityGovernance = consultationRequired`
6.5	Federated + community approval	`communityGovernance = communityApprovalRequired`
6.6	Hybrid public/restricted	distribution-level constraints

← Level 5

End of Level 6 Documentation — FAIR² Access Governance